X-ray absorption in relation to valency of iridium in sputtered iridium oxide films

Journal of Electroanalytical Chemistry 465 (1999) 88–95

X-ray absorption in relation to valency of iridium in sputterediridium oxide films

Thierry Pauporte a,*, Daniel Aberdam b, Jean-Louis Hazemann c, Rene Faure a,Robert Durand a

a Laboratoire d’Electrochimie et Physico-Chimie des Materiaux et des Interfaces, UMR INPG-CNRS 5631, Domaine Uni6ersitaire,1130 rue de la piscine, BP 75, 38402 Saint Martin d’Heres, France

b Laboratoire de Cristallographie, UPR CNRS 5031, 25 rue des martyrs, BP166X, 38042 Grenoble Cedex, Francec Laboratoire de Geophysique Interne et Techtonophysique, UMR UJF-CNRS 5559, BP 53, 38041 Grenoble Cedex 09, France

Received 15 October 1998; received in revised form 18 January 1999

Abstract

Electronic and structural changes induced by the charge storage reaction due to proton insertion in sputtered iridium oxide films(SIROFs) have been investigated by in situ X-ray absorption spectroscopy at the L3 edge of iridium atoms in 1 M H2SO4. Theiridium valency is shown to increase from 3 to 3.85 when the potential varies from −0.2 to +1 V(SCE). In XANES spectra, thewhite line peak height and energy position decrease with insertion. The fine structures of the spectra have been analyzed andsimulated in view of structural parameter extraction. A correspondence curve is established between the interatomic Ir–O distancein the first shell and the iridium valency. A strong decrease of this distance is observed with the oxidation state of iridiumaccompanied by a conspicuous decrease of the Debye–Waller factor. © 1999 Elsevier Science S.A. All rights reserved.

Keywords: Adsorption; Sputtered iridium oxide films; Proton insertion

1. Introduction

Thermal decomposition of iridium salts [1], anodicoxidation of a metal film [2,3], reactive sputtering in anO2 plasma [4,5] and anodic [6,7] or cathodic [8] elec-trodeposition are the main routes for obtaining iridiumoxide films (IROFs). Those obtained with the last threemethods exhibit an interesting electrochromic property:they are transparent in the reduced state and black inthe oxidized state. This behavior is the opposite of thatobserved with another largely studied electrochromiccompound, WO3. Therefore, the IROF has been sug-gested for use as a counter electrode in electrochromic

devices with a WO3 electrode to play the role of anenhancer for the color change [9]. To the best of ourknowledge, no electrochromism is observed in IROFsobtained from the classical thermal decompositionmethod which give rise to well crystallized films [10].

It is currently accepted that the mechanism of elec-trochromism in anodic iridium oxide films (AIROFs)and sputtered iridium oxide films (SIROFs) consists ofthe insertion of both electrons and cations in the film.In acid electrolytes (pHB4), several investigations byGottesfeld and McIntyre et al. with AIROFs favored aproton mechanism [2,3]. The involvement of protonswas confirmed later by direct elemental analysis ofAIROF composition by ex situ Rutherford back-scat-tering and nuclear reaction analysis techniques [11], andfrom the analysis of the variation of the electrolytedensity before and after insertion in AIROFs [12] andSIROFs [13,14] by probe beam deflection in combina-tion with cyclic voltammetry.

* Corresponding author. Present address: Laboratoire d’Elec-trochimie et Chimie Analytique, UMR CNRS 7575, ENSCP, 11 ruePierre et Marie Curie, 75231 Paris Cedex 05, France. Fax: +33-476-826777.

E-mail address: [email protected] (T. Pauporte)

0022-0728/99/$ - see front matter © 1999 Elsevier Science S.A. All rights reserved.PII: S 0 0 2 2 -0728 (99 )00058 -3

T. Pauporte et al. / Journal of Electroanalytical Chemistry 465 (1999) 88–95 89

The electrochromic charge storage reaction generallyassumed is:

IrO2+H+ +e−UIrO(OH) (1)

in SIROFs and:

IrO(OH)2+H+ +e−UIr(OH)3 (2)

in AIROFs. Some authors have suggested that Reac-tion 1 could occur in the bulk of AIROFs [11].

Ex situ (in air) EXAFS measurements at the L3

iridium edge with well crystallized iridium oxide [15,16]and with SIROFs [15,17] have been reported in theliterature. This technique is particularly well suited forprobing the local order of the oxygen octahedra aroundthe iridium atoms. The shell in SIROFs has beensimulated with two Ir–O distances and with fixing ofthe total number of oxygen (n) at 6. Distances rangingbetween 1.934 and 2.017 A have been found.

Huppauff and Lengeler [18] have reported in situXAS measurements with AIROFs (performed in anelectrochemical cell) at the L1 and L3 iridium edges.They have determined that the oxidation state of irid-ium in AIROFs (denoted 6) can vary from 3 to 4.8 withthe applied potential, indicating that a mixture of Ir4+

and Ir5+ is present at the higher potentials investigated(up to 1.2 V (SCE)). A strong decrease of the inter-atomic Ir–O distance (denoted r) accompanies the va-lence increase.

In the present work we have probed by in situ XAS,the changes induced by potential in a film obtained byreactive sputtering, actually the simpler and most usualmethod for obtaining electrochromic IROFs. The rela-tionship between the electrode potential and 6 has beenstudied in order to compare our results with those inRef. [18].

In Ref. [18], 6 was determined by comparing theposition in energy of the white line peak of the AIROFwith that obtained with two model compounds ofvalency +3 and +4, assuming a linear dependency ofthis energy upon 6. We have encountered two mainproblems with our results using the same procedure: theaccuracy of our white line position was relatively low(close to 0.5 eV as detailed below) and in contrast withRef. [18] our IrO2 reference was not pure Ir4+ but amixture of Ir3+ and Ir4+. To overcome thesedifficulties, we have used an alternative procedure fordetermining 6 in SIROFs. The first step consisted ofdetermining the potential at which 6 is equal to +3from the r versus electrode potential curve obtainedfrom EXAFS data analysis. Secondly, the charge storedby Reaction 1 as a function of electrode potential hasbeen measured by cyclic voltammetry. This method canbe used with SIROFs because these films are very stablein the medium studied and consequently, the quantityof iridium atoms they contain can be determined easilyand with good accuracy from the film thickness. Con-

versely, if AIROFs are stable in the medium, theiriridium atom quantity is not well known because apartial dissolution occurs during the oxidation by cy-cling of the as-deposited metallic iridium film. At themost positive potentials, formation of soluble oxidizediridium species is described in the literature [3]. Themethod used in Ref. [18] leads to a direct determinationof the absolute valency of the Ir atom in so far as theaccuracy on the white line position is good enough andas pure reference samples are available. In the resultspresented here it has been assumed that the valencestate of oxygen remains constant (= −2) and that onlythe voltammetric charge exchanged gives rise to the Irvalency change. To our knowledge, no polarizationeffect of the Ir–O bond in the bulk of the SIROF isdescribed in the literature. In Section 4 we show thatthe results obtained with the two procedures are ingood agreement. Different electrochemical behavior isfound between the AIROF and SIROF. With the lattercompound the results are interpreted with a singleredox Ir3+/Ir4+ process between −0.2 and +1.1 V(SCE).

2. Experimental

2.1. Samples and electrochemical cell

The reference compound used for IrO2 was the com-mercial iridium oxide hydrate of purity 99.9% fromAldrich. Before the study, the powder was heated at450°C for 2 h under an air flux in order to try andeliminate any residual water. The amount of powdercorresponding to ca. one absorption length was thenmelted with an appropriate quantity of sugar andpressed in order to obtained a pellet of several mm inthickness. X-ray absorption measurements were per-formed in transmission mode with this referencesample.

IrO2 thin films were prepared and kindly supplied bySaint Gobain Recherche. They were deposited by reac-tive RF sputtering from a metallic iridium target in anO2 plasma onto a porous carbon substrate of highpurity (carbon paper from Toray). Two film thicknesseswere available: 21 and 59 nm. These values were de-duced from the step obtained between a masked and anon-masked part of the glass plate supporting theToray sample during deposition and measured with aprofiler.

The electrochemical cell used was constructed usingthe filter press concept. The frame consisted of two ringplates 58 mm in outer diameter and 26 mm in innerdiameter made from duralumin. Between them, a 100mm thick kapton foil, the working electrode, a 3 mmthick polyethylene separator enclosing the electrolyticsolution, the counter electrode made from a sheet of

T. Pauporte et al. / Journal of Electroanalytical Chemistry 465 (1999) 88–9590

Toray paper and another kapton foil were placed suc-cessively. These different layers were maintained be-tween the two rings with six screws. The electricalcontacts on the Toray papers were taken with twostrips of gold foil. The active part of the workingelectrode was a disk 10 mm in diameter in contact withthe electrolytic solution. This cell was designed in orderto allow the X-ray beam to pass through. The incomingbeam illuminated the back of the working electrode.The Ka fluorescence signal was recorded at 90° with a30 element germanium solid state detector from theCanberra Company.

The electrolytic solution was 1 M H2SO4 preparedfrom concentrated Merck suprapur sulfuric acid andMillipore water. Flexible capillaries 1 mm in diameterwere connected to the separator compartment throughholes drilled in the bottom and the top of the separatorplate. They allowed the filling of the cell and the flow ofsolution with the help of a peristaltic pump. The pumpwas stopped during the recording of spectra. The satu-rated calomel electrode (SCE) reference electrode wasplaced in the solution circuit, upstream from the elec-trochemical cell. In the following, all potentials arereferred to the SCE. The potential of the cell wascontrolled with an EG&G PAR 273A potentiostat.

Results obtained with four different samples are pre-sented. Sample A was a 21 nm thick SIROF studied inthe electrochemical cell. Sample B was an as-deposited21 nm thick SIROF studied in air. Its potential wasmeasured immediately after immersion in an electro-chemical cell, and found to be equal to 0.35 V. SamplesC and D were 59 nm thick SIROFs studied in situ. The59 nm thick samples were pretreated before XASrecordings. The pretreatment consisted of 50 repetitivepotential cyclings of the electrodes between 0 and 1 Vat 100 mV s−1. In contrast to what has been observedby electrochemical impedance spectroscopy [19], no sig-nificant change attributable to the pretreatment hasbeen observed from our XAS experiments showing thatit induces no significant local change.

2.2. XAS measurements and EXAFS spectra analysis

The X-ray absorption experiments were conducted atthe B32 beam line (Collaborative Research Group onSurface and Interface (CRG-IF)) at the European Syn-chrotron Radiation Facility (ESRF) in Grenoble. Theelectron energy in the storage ring was 6 GeV with anoperating current ranging between 160 and 200 mA.The X-ray beam was monochromatized with a doublecrystal of Si(111). Harmonic rejection was performedusing a nickel-coated mirror. The beam size was 0.3×0.3 mm2 at full-width half-maximum and the photonflux on the sample ca. 1011 s−1. The energy resolutionof the monochromator, DE/E, was 2×10−4. All mea-surements have been performed at the iridium L3 edge

located at ca. 11 215 eV. The XAS spectra wererecorded from 11 000 to 12 730 eV (reference) or 12 065eV (SIROFs).

Generally, several spectra have been summed beforeanalysis using a cumulative approach. The principles ofEXAFS spectra extraction and simulation are describedin Ref. [20]. The analysis was performed using softwarewritten by one of us (D. Aberdam) [21]. After a classi-cal background removal, the EXAFS oscillations x(k)were multiplied by k3 in order to minimize the weightof the low k values where multiple scattering may playan important role. In order to minimize truncationeffects, a Kaiser window with parameter t=2 wasapplied prior to Fourier transforming, between k=1.7and 13 A−1 (between 1.7 and 20 A−1 for the referencespectrum). The first shell was then isolated between ca.0.9 and 2.3 A and its contribution to the signal recov-ered by an inverse Fourier transformation.

The three parameters which describe the electronmean-free-path G, h, b [21] and the prefactor S0

2, havebeen determined from the best fitting of the k3x(k)spectrum of the first shell of the IrO2 reference usingMacKale’s tables. They have been kept fixed at 1.67,5.05, 0.209 and 0.80, respectively, for SIROF spectrumfittings.

The reference sample has been used as a model fromwhich backscattering amplitude and phase shift fileshave been extracted. These files have been used asreferences to determine the structural parameters of theEXAFS formula from successive phase and amplitudedecorrelations. The range of wave numbers for simula-tion was 3–8 A−1 with 21 nm thick samples (A and B)and 3–11 A−1 with 59 nm thick samples (C and D).

3. Results

3.1. Electrochemistry

Fig. 1 shows a typical cyclic voltammogram recordedbetween −0.2 and +1.1 V with a 21 nm thick SIROFdeposited on Toray carbon. Under identical conditionsthe electrochemical response of the carbon substrate(Toray paper) was much lower in amplitude and thevoltammogram of Fig. 1 is due only to the film re-sponse. At the lower scan rates, the current is propor-tional to the sweep rate and it has been checked thatthe current was proportional to the SIROF thickness.The voltammetric charge between −0.2 and +1.05 Vwas large and equal to 8.2 mC cm−2. Two bumps arepresent at 0.55 and 0.38 V during the positive andnegative going scans, respectively. We have observedthat films grown on fluoride-doped SnO2 deposited onglass panes present two well defined symmetrical peaksat ca. 0.52 V [19]. With these samples, the voltammetriccharge was close to that found here. Their surfaces were


very smooth with no features observed by scanningelectron microscopy above 0.03 mm. SIROFs are notporous and the simplest explanation for the proportion-ality between the charge and the film thickness is thatelectrochemical insertion of protons occurs by Reaction1. This is argued for by the proportionality between theoptical density (bulk phenomena) and the voltammetriccharge reported in Ref. [5]. The shape of the voltam-mograms depends on the method used to prepare thefilms and a quite different response is obtained withporous AIROFs [2,3,18,22]: the major anodic peak is at0.75 V and the cathodic peak at 0.70 V; a small anodicpeak is also observed at 0.42 V. At potentials lowerthan 0.3 V a low capacitive behavior due to the doublelayer is observed. Finally, the anodic peak observed inFig. 1 above 1.05 V is attributable to the onset of theoxygen evolution reaction.

3.2. X-ray absorption spectra

Fig. 2 shows the typical variations of the near edgeprofile (XANES) we observed with the change of ap-plied potential. A noticeable shift of 1.2 eV of theenergy position of the white line peak is observedbetween 0 and 1 V. The shift is accompanied by anincrease of the white line height. Thus, the XANESspectra show clearly that the films respond readily tothe applied potential. When several spectra wererecorded successively under identical conditions, slightvariations of the peak position were observed. Thegreatest variations were 0.5 eV and give a good estima-tion of the uncertainty of the energy determinationduring our experiments.

The fine structure x(k) of the spectra is now consid-ered. Fig. 3 shows the Fourier transforms of k3x(k)spectra without phase shift correction for the IrO2

Fig. 2. Variation of the near edge profile of a SIROF versus theelectrode potential (59 nm thick sample D).

reference sample and for one of the SIROFs. Thereference curve presents a first peak centered at 1.57 Awhich corresponds to the backscattering of the firstoctahedral shell. The second main peak is of similarintensity with the weighting window used, centered at3.44 A and attributed to the backscattering of the firstshell of iridium neighbors [15]. The EXAFS spectrumconfirms that the reference compound is rather wellcrystallized. If we compare the reference spectrum withthat recorded with a SIROF using the same weightingwindow, the first peak is present with a similar ampli-tude whereas the second peak has a much lower ampli-tude. This can be related to the amorphous structure ofSIROFs [15]. XAS is a particularly accurate techniqueto probe the local order in this kind of compound.

Fig. 3. Fourier transform of k3x(k) without phase correction. (a)Dashed line: reference IrO2 sample (weighting window 3–20 A−1);(b) solid line: 59 nm thick SIROF (sample C at 1 V, weightingwindow 3–13 A−1).

Fig. 1. Voltammogram of a 21 nm thick SIROF deposited on Toraypaper measured at a scan rate of 10 mV s−1 in a 1 M H2SO4

solution.


Fig. 4. Fourier filtered contribution of the first shell due to oxygenneighbors of the spectrum presented in Fig. 3 curve b. Solid line:experimental curve; dashed line: simulated curve.

Fig. 6. Variation of the number of closer neighbors versus theelectrode potential in SIROFs.

that observed with the thicker ones.Fig. 5 shows the variation of Ir–O distance in the

first shell. The dispersion of the experimental pointscorresponds to an error in r determination of ca. 2×10−2 A, within the typical errors for the technique. Inspite of this dispersion a conspicuous decrease of r from2.01 down to 1.97 is observed when the applied poten-tial is increased from 0 to 1 V.

In Fig. 6 the variation of the number of closeroxygen neighbors versus the applied potential is pre-sented. No significant variation is observed between−0.1 and +1 V. The short range order in the film isassumed to be of the rutile type and then n is expectedto be equal to 6. The values determined from simula-tions are closer to ca. 6.3 but the uncertainties on thenumber of neighbors are too large (close to 10% classi-cally) to state that n is different from 6. This parameterdoes not seem to vary with the potential.

Fig. 4 shows a typical curve obtained from Fourierfiltered k3x(k) spectra. The corresponding curve fitobtained from simulation is presented in a dashed line.Figs. 5–8 show the parameters calculated from the firstshell fitting for the different samples studied as a func-tion of electrode potential. The parameters determinedat 1.2 V do not seem to reflect the variations observedat other potentials probed from −0.1 to +1 V. At thisformer potential, oxygen evolution reaction occurs andgas formation could perturb the recording of XASspectra. Similarly, we had observed perturbations inXAS spectra when the pump was not stopped duringthe spectrum recordings. Consequently, the followingdiscussion does not take into account the parametersdetermined at 1.2 V. The two other points obtainedwith 21 nm thick samples show a behavior similar to

Fig. 5. Variation of the distance between iridium and oxygen atoms inthe first shell versus the electrode potential in SIROFs. Fig. 7. Energy shift versus the electrode potential in SIROFs.


Fig. 8. Debye–Waller factor versus the electrode potential inSIROFs.

charge exchanged goes into the film (if any superficialeffects exist, they are neglected), we have constructedthe Ir valency versus potential curve using the averagevalue of the exchanged charge between the positive andnegative going scans of Fig. 1. The quantity of Ir atomsin the films has been calculated assuming that thedensity of the as-deposited SIROF is equal to 10 gcm−3 [5] and its molar mass is equal to 224 g mol−1.The curve of correspondence is presented in Fig. 9showing behavior similar to that obtained with samplesgrown on fluorine-doped SnO2 [19] but different fromthat obtained with AIROFs [18] in which the 6 changestarts at 0.3 V and then increases more drastically withE.

From Fig. 9, it can be estimated that the shift of thewhite line peak presented in Fig. 2 is of the same orderof magnitude as that observed with an AIROF in so faras the uncertainty of our energy measurement has beenmeasured close to 0.5 eV. A 1.2 eV shift for D6 equal to0.8 is found with the films studied here whereas it wasof ca. 2 eV with a 6 change of 1.8 with AIROFs [18].

Variations of the distance r, are reported as a func-tion of 6 in Fig. 10. In spite of the dispersion, aquasi-linear variation is found and represented with thesolid line of regression. In the same figure we have alsopresented, with a dashed curve, the data obtained withAIROFs and extracted from Ref. [18] for comparison.It presents two quasi-linear parts with a break at 6equal to ca. 3.6. The two curves are globally in goodagreement showing the consistency of the two proce-dures adopted for constructing the r versus E curves.

Information on the iridium ionic radii can be de-duced from our data. We assume that the trivalentform of the SIROFs studied is IrO(OH) and the te-travalent form is IrO2 and we know that O2− and(OH)− radii are 1.38 and 1.35 A, respectively [23].Then Ir3+ and Ir4+ radii are found equal to 0.645 and

Fig. 7 shows the variations of the energy shift withpotential. Globally, a slight increase is observed mainlydue to the variation of the edge energy with potential.For instance a shift of ca. 2 eV is found with samples Dbetween 0 and 1 V, in rather good agreement with thevariation of the energy position of the white line peakobserved in Fig. 2. In the EXAFS formula, the ampli-tude is damped by an exp(−2s2k2) factor. The De-bye–Waller factor, s2, is related to the standarddeviation s, around the distance r, and is thus a mea-sure of the disorder in the shell. Fig. 8 shows a signifi-cant decrease of this factor with increasing electrodepotential.

4. Discussion

As explained in Section 3.1, the charge storage prop-erties of IROFs depend on the method of preparation.It is thus impossible to compare the results given in theprevious section as a function of the applied potentialwith those published for AIROFs. Beforehand, thecorrespondence between electrode potential and iridiumvalency must be established. The method used is nowdescribed briefly. From structural data of trivalent irid-ium oxalate, a compound with six oxygen atoms asnearest neighbors in an octahedral configuration, Hup-pauff and Lengeler [18] have calculated that the dis-tance between an iridium atom of valency +3 (thelowest stable non-zero valency of iridium) and an oxy-gen atom must be close to 2.02 A. Fig. 5 shows thatwith our SIROF samples this distance is fortuitouslyobtained at a potential close to that of hydrogen evolu-tion. Consequently, we have supposed that the oxida-tion state of Ir is equal to +3 at −0.2 V, a potentialconveniently located before the onset of hydrogen evo-lution, as displayed in Fig. 1. Assuming that all the Fig. 9. Iridium valency versus the electrode potential in SIROFs.


Fig. 10. Iridium–oxygen bond length variation versus iridium valencyin SIROFs. Solid line: line of regression; dashed line: experimentalcurve with AIROFs extracted from Ref. [18].

5. Conclusions

SIROFs have been investigated at the L3 edge ofiridium by XAS in an electrochemical cell. The set uphas been constructed to record spectra at controlledpotential. XANES shows an increase in the white linepeak energy and height with electrode potential due tothe change in iridium valency. The short range order inthe octahedral first shell of oxygen atoms has beeninvestigated. A strong decrease of the Ir–O distancefrom 2.011 to 1.963 A has been found with increasingiridium oxidation state from 3.03 to 3.85. These resultshave been shown to be in good agreement with thoseobtained with AIROFs. Related to the decrease of r, adiminution of the Debye–Waller coefficient has beenshown and attributed to the effect of r on the dynamiccontribution of this parameter. Finally, we have shownthat the higher oxidation state of iridium in SIROFsbefore oxygen evolution is close to +4 and not +4.8as found with AIROFs.

Acknowledgements

The authors wish to acknowledge C. Lefrou, whosupplied them with SIROF samples when she was atSaint-Gobain Recherche. Y. Soldo and the technicalstaff of the IF spectrometer are acknowledged for theirkind help during XAS experiments at the ESRF.

References

[1] S. Trasatti, in: J. Lipkowski, P.N. Ross (Eds.), The Electrochem-istry of Novel Material, Ch. 5: Transition Metal Oxides: Ver-satile Materials for Electrocatalysis, VCH, New York, 1994, p.207.

[2] S. Gottesfeld, J.D.E. McIntyre, G. Beni, J.L. Shay, Appl. Phys.Lett. 33 (1978) 208.

[3] S. Gottesfeld, J.D.E. McIntyre, J. Electrochem. Soc. 126 (1979)742.

[4] L.M. Schiavone, W.C. Dautremont-Smith, G. Beni, J.L. Shay,Appl. Phys. Lett. 35 (1979) 823.

[5] L.M. Schiavone, W.C. Dautremont-Smith, G. Beni, J.L. Shay, J.Electrochem. Soc. 128 (1981) 1339.

[6] K. Yamanaka, Jpn. J. Appl. Phys. 28 (1989) 632.[7] M. Petit, V. Plichon, J. Electroanal. Chem. 444 (1998) 247.[8] J.A. Cox, S.E. Gadd, B.K. Das, J. Electroanal. Chem. 256 (1988)

199.[9] C. Desporte, Electrochimie des Solides, Presses Universitaires de

Grenoble, Grenoble, 1994, p. 420.[10] F. Andolfatto, Ph.D. thesis, Institut National Polytechnique de

Grenoble, Grenoble, 1992.[11] J.D.E. McIntyre, S. Basu, W.F. Peck, W.L. Brown, W.M.

Augustyniak, Phys. Rev. B 25 (1982) 7242.[12] R. Kotz, C. Barbero, O. Haas, J. Electroanal. Chem. 296 (1990)

37.[13] M. Petit, V. Plichon, J. Electroanal. Chem. 379 (1994) 165.[14] M. Bardin, P. Loheac, M. Petit, V. Plichon, N. Richard, New J.

Chem. 19 (1995) 59.

0.575 A, respectively, using the regression curve of Fig.10.

In contrast to the results reported with AIROFs, wehave not found the presence of Ir with valence +5 overthe potential range studied. At 1.1 V the valency isclose to +4. At higher potentials, at which the oxygenevolution reaction occurs, Ir atoms could start to oxi-dize at a higher valency state and play an importantrole in electrocatalysis.

The interatomic distance obtained from the analysisof the EXAFS spectrum of the IrO2 reference com-pound is 1.977 A. From the regression curve of Fig. 10,the valency of the iridium is determined equal to 3.58.This illustrates the observation made above about thedifficulty of obtaining a model containing only Ir4+

and not a mixture of Ir3+ and Ir4+. On the other hand,the IrO2 reference sample used by Huppauff andLengeler [18] turns out most likely to be pure Ir4+

because the results obtained here and those presentedby Huppauff and Lengeler are in good agreement.

The Debye–Waller factor contains two contributionsdue to dynamic (thermal) and static disorders [20]. Thedynamic disorder usually increases with bond lengthand the decrease observed in Fig. 8 with potential islikely to be due to the decrease of r. The main valuesobtained are significantly larger than that found forcrystallized IrO2. They vary from 7×10−3 to 5.5×10−3 A2 when the potential increases from 0 to 1 V.These values correspond to a spread of interatomicdistances ranging from 0.2 to 0.17 A (FWHM) resp.The Debye–Waller parameter calculated for the wellcrystallized IrO2 reference, 4.36×10−3 A2, turns out tobe significantly smaller than that found for SIROFs ofthe same 6. This shows a higher degree of static disor-der in SIROFs which is likely to be because they aremainly amorphous.


[15] N. Bestaoui, E. Prouzet, P. Deniard, R. Brec, Thin Solid Films235 (1993) 35.

[16] E. Prouzet, J. Phys. Condens. Matter 7 (1995) 8027.[17] A. Barlerna, E. Bernieri, E. Burattini, A. Kuzmin, A. Lussis, J.

Purans, P. Cikmach, Nucl. Instrum. Methods Phys. Res. A308(1991) 234.

[18] M. Huppauff, B. Lengeler, J. Electrochem. Soc. 140 (1993) 598.

[19] T. Pauporte, R. Durand, submitted for publication.[20] D.C. Koeningsberger, R. Prins, X-ray Absorption: Principles,

Applications, Techniques of EXAFS, SEXAFS and XANES,Wiley, New York, 1988.

[21] D. Aberdam, J. Synchrotron Radiat. 5 (1998) 1287.[22] D.A.J. Rand, R. Woods, J. Electrochem. Soc. 55 (1974) 375.[23] R.D. Shannon, Acta Crystallogr. A 32 (1976) 752.

..

Documents

X-ray absorption in relation to valency of iridium in sputtered iridium oxide films